Uncertainty, slow government and optimal taxes

The optimal taxation literature has come up with incredibly complex and non-linear taxation schemes that have no way of making it into policy. The most complex tax code, the US one, is not complex because that would be socially optimal, it is complex because of special interests. Complex code does not make it past the politicians because they do not understand it, because it appears not to be transparent, or because its mathematical complexity scares everyone.

Marcus Berliant and Shota Fujishima claim rather that the lack of complexity on the tax code has to do with sluggishness. Governments can simply not adapt the tax code as fast as conditions change. Of course, they could also set up contingent rules, but I suppose that this is deemed opaque and subject to interpretation. Anyway, one consequence of this sluggishness is that optimal taxes start looking very different. A typical result of standard theory is that the top earner should have a zero marginal tax rate, so as to encourage this most productive person to work more. But when this person changes from period to period, or if this is the same person but he does not reach the maximum income every period, the result does not carry through. The top marginal tax rate needs to be positive to leave some headroom. But the marginal tax rate is still declining at the top. In a way, this is achieved in many countries by allowing for "loopholes" that makes it possible for the most productive people to pay less taxes.